Dynamic Regulation of the Cerebral Cavernous Malformation Pathway Controls Vascular Stability and...

Developmental Cell

Article

Dynamic Regulation of the CerebralCavernous Malformation PathwayControls Vascular Stability and GrowthXiangjian Zheng,1 Chong Xu,1 Annie O. Smith,4 Amber N. Stratman,4 Zhiying Zou,1 Benjamin Kleaveland,5 Lijun Yuan,1,6

Chuka Didiku,1 Aslihan Sen,1 Xi Liu,1,6 Nicolas Skuli,2,3 Alexander Zaslavsky,3 Mei Chen,1 Lan Cheng,1 George E. Davis,4

and Mark L. Kahn1,*1Department of Medicine and Cardiovascular Institute2Howard Hughes Medical Institute3Abramson Family Cancer Research Institute

University of Pennsylvania, Philadelphia, PA 19104, USA4Department of Medical Pharmacology and Physiology, School of Medicine, University of Missouri-Columbia, Columbia, MO 65212, USA5Department of Pathology, Massachusetts General Hospital, Boston, MA 02114, USA6Fourth Military Medical University, Xian 710032, China

*Correspondence: [email protected]

http://dx.doi.org/10.1016/j.devcel.2012.06.004

SUMMARY

Cardiovascular growth must balance stabilizingsignals required to maintain endothelial connectionsand network integrity with destabilizing signalsthat enable individual endothelial cells to migrateand proliferate. The cerebral cavernous malforma-tion (CCM) signaling pathway utilizes the adaptorprotein CCM2 to strengthen endothelial cell junctionsand stabilize vessels. Here we identify a CCM2paralog, CCM2L, that is expressed selectively inendothelial cells during periods of active cardiovas-cular growth. CCM2L competitively blocks CCM2-mediated stabilizing signals biochemically, incultured endothelial cells, and in developing mice.Loss of CCM2L reduces endocardial growth factorexpression and impairs tumor growth and woundhealing. Our studies identify CCM2L as a molecularmechanism by which endothelial cells coordinatelyregulate vessel stability and growth during cardio-vascular development, as well as postnatal vesselgrowth.

INTRODUCTION

The heart and blood vessels mediate gas exchange and deliver

nutrients, signalingmolecules, and circulating cells to the tissues

of the body. The vertebrate cardiovascular system is lined by

specialized endothelial cells that direct its growth and function.

During cardiovascular development the heart and vessels are

first formed by endothelial cells that arise from mesodermal

precursors through a process of cardiogenesis and vasculogen-

esis (Potente et al., 2011; Risau, 1997). After de novo formation

of the heart and earliest embryonic vessels, vascular growth

occurs through angiogenic sprouting of endothelial cells from

preexisting vessels (Potente et al., 2011).

342 Developmental Cell 23, 342–355, August 14, 2012 ª2012 Elsevie

In the functioning cardiovascular system, endothelial cells

must be tightly connected to each other through cell-cell junc-

tions to maintain a closed vascular network through which blood

can circulate (Dejana et al., 2009). In contrast, during angiogen-

esis endothelial cells must transiently disconnect from each

other and the existing network in order to proliferate andmigrate.

Endothelial cell junctions and vessel stability must therefore be

molecularly regulated during vascular growth in a highly spatially

and temporally coordinated manner to allow growth without

compromising the integrity of the existing cardiovascular

network. Vascular endothelial growth factor (VEGF), a protein

that both loosens endothelial junctions and stimulates endothe-

lial proliferation (Murohara et al., 1998; Senger et al., 1983), is one

such regulator. However, because tumor vessels are able to

overcome the effects of VEGF blockade, other molecular mech-

anisms of regulating vessel stability and vessel growth must

exist, and their identification is critical to design more effective

therapies.

The cerebral cavernous malformation (CCM) signaling

pathway has recently been identified as a critical positive regu-

lator of endothelial junctions and vessel stability. The CCM

pathway consists of three adaptor proteins, KRIT1 (aka

CCM1), CCM2, and PDCD10 (aka CCM3) that were identified

as disease genes in patients with cerebral vascular malforma-

tions. The CCM proteins bind each other (Voss et al., 2007)

and the HEG receptor (Kleaveland et al., 2009). Human CCMs

exhibit defective endothelial junctions (Clatterbuck et al.,

2001), and loss of HEG, CCM1, CCM2, or CCM3 function results

in abnormal endothelial cell junctions and vascular lumen forma-

tion in mice and zebrafish in vivo and endothelial cells in vitro

(Glading et al., 2007; Kleaveland et al., 2009; Stockton et al.,

2010;Whitehead et al., 2009; Zheng et al., 2010). Genetic studies

in mice and fish have also demonstrated that the CCM signaling

pathway plays an essential and conserved role in cardiovascular

development. Mice and fish lacking CCM1 or CCM2 fail to

develop lumenized branchial arch arteries that connect the heart

to the aorta (Whitehead et al., 2004, 2009; Zheng et al., 2010),

and loss of HEG in both species confers defects in heart growth

(Kleaveland et al., 2009; Mably et al., 2003, 2006). We

r Inc.

mailto:[email protected]


Developmental Cell

CCM2L Controls Vascular Stability and Growth

hypothesized that the HEG-CCM signaling pathway must be

regulated to permit efficient angiogenesis and cardiogenesis.

To test this hypothesis we searched for novel regulators of

angiogenesis that might function through the CCM pathway.

We demonstrate that a paralog of CCM2 (CCM2L) opposes the

stabilizing effects of CCM signaling to liberate angiogenic endo-

thelial cells during cardiovascular growth. Biochemical studies

and genetic analysis of mice lacking HEG, CCM2, and/or

CCM2L reveal that CCM2L functions by competing with CCM2

for binding to the HEG-CCM1 complex and uncoupling these

upstream components of the pathway from CCM3, a critical

stability effector, while activating expression of factors that

support cardiovascular growth. Loss of CCM2L prevents tumor

growth in mice and delays wound healing, findings consistent

with a specific role in regulating angiogenesis in vivo. Ccm2LGFP

reporter mice reveal that CCM2L expression in vivo is detected

only in endothelial cells, especially those that participate in active

cardiovascular growth. We propose that CCM2L functions as

a molecular mechanism through which CCM signaling converts

endothelial cells from a stable to an angiogenic phenotype and

by which endothelial responses in vascular disease and growth

may be specifically targeted.

RESULTS

Identification of a CCM2 Paralog that Binds CCM1and HEG but Not CCM3To identify potential novel regulators of the CCM signaling

pathway, we used BLAST searching of the EST and Ensemble

databases to identify genes encoding structurally related pro-

teins.Agene that encodes aprotein predicted tobehighly homol-

ogous toCCM2 thatwedesignatedCCM2L (akaBC020535 in the

mouse and C20orf160 in the human) was identified in the human,

mouse, and zebrafish databases (Figure 1A; Figure S1 available

online). The best-characterized domain of the CCM2 protein is

the phosphotyrosine binding (PTB) domain that mediates inter-

action with CCM1 (Zawistowski et al., 2005). Ccm2L encodes

a PTB domain that contains a long insertion between the b6

and b7 sheets but is highly homologous to that of CCM2 in the

b5 strand through which DAB1-like PTB domains are believed

to interact with peptide ligands (Figure 1A; Stolt et al., 2003; Uhlik

et al., 2005), suggesting that CCM2L may also bind CCM1.

Biochemical studies have demonstrated that CCM2 binds

CCM1 via its PTB domain, and CCM3 through an as yet unde-

fined region of the protein, to nucleate a signaling complex that

is recruited to the HEG receptor by CCM1 (Kleaveland et al.,

2009; Voss et al., 2009; Zawistowski et al., 2005; Zheng et al.,

2010). To determine if CCM2L proteins are also able to complex

with CCM1 and CCM3, epitope-tagged CCM2 and CCM2L

proteins were coexpressed with CCM1 and/or CCM3 in

HEK293T cells and coimmunoprecipitation experiments were

performed. CCM1 coimmunoprecipitated efficiently with both

CCM2 and CCM2L (Figure 1B). In contrast, CCM3 coimmuno-

precipitated with CCM2 and not with CCM2L (Figure 1C).

Studies of CCM2 binding to CCM1 have identified two PTB

domain point mutants, L198R and F217A, that disrupt binding

to CCM1 (Zawistowski et al., 2005). To further investigate the

mechanism by which CCM2L binds CCM1, we next tested

whether the equivalent CCM2L PTB domain point mutants,

Develop

L306R and F325A (see Figure 1A for comparison to mouse

sequence in which L198 is M198), are capable of binding

CCM1. The L306R point mutant that is equivalent to a CCM

disease-associated mutation in CCM2 (Denier et al., 2004), but

not the F325A mutant, conferred severe loss of CCM1 binding

by CCM2L (Figure 1D). Thus, CCM2L binds CCM1 via its PTB

domain in a manner that is similar but not identical to that of

CCM2. Previous studies have demonstrated that the CCM

protein complex associates highly efficiently with the HEG

receptor intracellular tail (HEG-IC) via CCM1 (Kleaveland et al.,

2009; Zheng et al., 2010). Beads coupled to the intracellular

(IC) tail of HEG but not integrin aIIb efficiently pulled down

a complex of ccm1, ccm2, and ccm3 when those proteins

were coexpressed in HEK293T cells (Figure 1E). In contrast,

when ccm1, ccm2L, and ccm3 proteins were coexpressed,

HEG-IC beads pulled down a complex containing only ccm1

and ccm2L (Figure 1E). These findings indicate that both

CCM2 and CCM2L interact with CCM1 and HEG via their PTB

domains but that ccm2L does not associate with ccm3, a critical

downstream effector of the known CCM signaling pathway.

CCM2L Competes with CCM2 for CCM1 BindingThe finding that CCM2L binds CCM1 suggested that CCM2 and

CCM2L may compete for CCM1 binding. Because endothelial

CCM1 and CCM2 levels were too low to detect endogenous

protein interaction in cultured endothelial cells and anti-CCM2L

antibodies are not yet available to determine if CCM2L and

CCM2 compete for binding to CCM1, we compared the ability

of known amounts of FLAG-taggedCCM2orCCM2L to compete

with HA-tagged CCM2 (Figure 1F) or CCM2L (Figure 1G) for

binding to CCM1. The addition of FLAG-CCM2L or FLAG-

CCM2 resulted in a dramatic reduction in the amount of CCM1

associated with HA-CCM2 (Figure 1F). Conversely, the addition

of FLAG-CCM2or FLAG-CCM2L resulted in adramatic reduction

in the amount of CCM1 associated with HA-CCM2L (Figure 1G).

Finally, addition of CCM2L L306R, a PTB point mutant with

severely reduced CCM1 binding (Figure 1C), failed to competi-

tively block CCM2 binding to CCM1 (Figure 1H). These studies

demonstrate that CCM2 and CCM2L compete for binding to

CCM1 and that the expression of CCM2L may thereby prevent

HEG-CCM1 complexes from associating with CCM3.

Ccm2L Expression Is Spatially Restricted to EndothelialCells and Temporally Linked to Cardiovascular GrowthAlthough an essential function for CCM2 has been demonstrated

in endothelial cells using genetic approaches (Boulday et al.,

2009; Whitehead et al., 2009), endothelial Ccm2 levels are too

low to be detected above background using either LacZ or

GFP knockin reporter alleles or radioactive in situ hybridization

(Kleaveland et al., 2009; Whitehead et al., 2009). Northern blot

analysis of adult mouse tissues revealed Ccm2L mRNA expres-

sion in adult heart and lung, organs with large endothelial cell

populations (Figure S1). To better define the expression pattern

of Ccm2L, we performed in situ hybridization studies using

mouse embryos from E9.5 to E18. Ccm2L mRNA was detected

exclusively along the endothelial cell border of the developing

heart and vessels (Figures 2A–2E).Ccm2L expressionwas stron-

gest in the E10.5 heart but was not detected in the heart or else-

where after E12.5 using in situ hybridization. Consistent with

mental Cell 23, 342–355, August 14, 2012 ª2012 Elsevier Inc. 343

Transfection:B

C

HA-CCM1 + + + + + + Flag-CCM2L - - + - - +Flag-CCM2 - + - - + -

Transfection:Myc-ccm3 + + + +

- +

HA-ccm2L - + - +HA-ccm2 + - + -

Flag-ccm1

HA-ccm2L

HA-ccm2

Myc-ccm3

D

IB:

A

E

Flag-CCM2 - - +Flag-CCM2L - + -

Myc-CCM1 + + +

Myc-CCM1

Lysa

te

F

Flag-CCM2L

Flag-CCM2

HA-CCM2 + + +

IB:

HA-CCM2IP: H

A

G

Myc-CCM1

Lysa

te Flag-CCM2L

Flag-CCM2

IB:

HA-CCM2LIP: H

A

HA-CCM1

Flag-CCM2LIB:

Flag-CCM2

5% InputIP: Flag

- ------- - - ++++++++++++ + +++++

Myc-ccm3

HA-ccm2LIB:

HA-ccm2

IP: HA 4% Input

HA-CCM1

HA-CCM1 + + + + + Flag-CCM2L - + - - -CCM2L-L306R - - + - -

5% In

put

IP: F

lag

CCM2L-LR/FA - - - - + CCM2L-F325A - - - + -

*

*

Flag-CCM2Ls

HA-CCM1

Flag-CCM2Ls

H

Flag-CCM2L - + -CCM2L-L306R - - +

Myc-CCM1 + + +

Myc-CCM1

Lysate

Flag-CCM2Ls

HA-CCM2 + + +

IB:

HA-CCM2IP: H

A

Flag-CCM2 - + -Flag-CCM2L - - +

Myc-CCM1 + + +HA-CCM2L + + +

Transfection:Flag-ccm1 + + + + + HA-ccm2L - - + - +HA-ccm2 + + - + -Myc-ccm3 + + + + +

aIIb HegIC

Bead pull-down

5% Input

*

Figure 1. CCM2L Competes with CCM2 for Binding to CCM1 but Does Not Bind CCM3(A) ClustalW alignment of the predicted PTB domains ofmouse CCM2L, CCM2, and DAB1 are shown. Green shading indicates identity in all three proteins, yellow

shading indicates identity in at least two proteins, and blue shading indicates conserved residues in at least two proteins. Red asterisks indicate CCM2L L306

and F325 amino acid residues. The predicted beta strands and alpha helices are ordered according to those of DAB1 PTB domain.

(B) CCM2L binds CCM1. Flag-CCM2 or Flag-CCM2L was coexpressed with HA-CCM1 in HEK293T cells, and anti-Flag immunoprecipitations (IP) were

performed.

(C) CCM2L does not bind CCM3. HA-ccm2 or HA-ccm2L was coexpressed with Myc-ccm3 in HEK293T cells, and anti-HA immunoprecipitations (IP) were

performed. Red asterisk notes a weak nonspecific detection of the IgG band.

(D) The CCM2L L306R, but not the CCM2L F325A, mutation disrupts CCM2L-CCM1 interaction. Flag-tagged CCM2L, CCM2L-L306R, CCM2L-F325A, or

CCM2L-L306R/F325A was coexpressed with HA-CCM1, and anti-Flag immunoprecipitations (IP) were performed.

(E) The HEG intracellular tail forms a complex with CCM1 and CCM2L that does not include CCM3. HA-ccm2L or HA-ccm2, FLAG-ccm1, and Myc-ccm3 were

expressed in HEK293 cells and pull downs performed with affinity matrices containing the intracellular tail of either the aIIb integrin subunit (aIIb) or the HEG

receptor (HegIC). Total protein expression in pull-down input is shown by immunoblot (IB) analysis (right).

(F–H) CCM2L and CCM2 compete for CCM1 binding. (F) Cell lysates containing the same amount of MYC-CCM1 and HA-CCM2 were mixed

with control lysate or lysate containing FLAG-CCM2L (middle lane) or FLAG-CCM2 (right lane) and HA-CCM2 immunoprecipitated with anti-HA antibodies.

The amounts of immunoprecipitated HA-CCM2 and coimmunoprecipitated MYC-CCM1 are shown above, and the amounts of FLAG-CCM2L and

Developmental Cell


344 Developmental Cell 23, 342–355, August 14, 2012 ª2012 Elsevier Inc.

Developmental Cell


these findings in the mouse, RT-PCR detected CCM2L expres-

sion in primary human microvascular endothelial cells and

primary human lymphatic endothelial cells but not in primary

human keratinocytes (Figure 2F). In contrast to CCM2L and

TIE2, a known endothelial specific gene, CCM1, CCM2, and

CCM3 were detected in equal abundance in primary keratino-

cytes and endothelial cells. Measurement of transcript copy

numbers revealed that CCM2L is expressed at levels similar to

those ofCCM1,CCM2, andCCM3 in endothelial cells (Figure 2F).

These studies suggest that CCM2L is expressed in an endothe-

lial-specific pattern and at a basal level that is similar to those of

CCM1, CCM2, and CCM3.

To further characterizeCcm2L expression and function in vivo,

we used homologous recombination in ES cells to place a cDNA-

encoding nuclear-localized GFP in frame with the start methio-

nine of Ccm2L (Figure S1). GFP was detected exclusively in

the nuclei of PECAM+ endocardial and endothelial cells of

Ccm2LGFP mice (Figures 2G–2T). GFP was not detected in all

endothelial cells but was instead found in a small subset of endo-

thelial cells that varied with embryonic age. Ccm2LGFP was

strongly expressed in the endocardium, where nuclear GFP

was detected by E9.5, peaked at E11.5 (when 20% of endocar-

dial cell nuclei were GFP+), and declined thereafter (Figures

2G–2K). Ccm2LGFP expression was undetectable in myocardial

cells at all time points (Figures 2G–2J). Ccm2LGFP expression

was strongly detected in the endothelial cells of the dorsal aorta

and cardinal vein prior to E12.5 but not at later time points

(Figures 2L–2N). Throughout gestation Ccm2LGFP expression

was detected in a small number of endothelial cells in vessels

scattered throughout the embryo (Figure 2O). These cells were

frequently adjacent to one another within a single vessel and

often found in nonlumenized extensions from lumenized,

blood-containing vessels (Figure 2O), for example, endothelial

cells invading the neural tube at E10.5 (Figures 2P and 2Q),

consistent with a role in active vessel and heart growth.

A site of highly active angiogenesis is the retina, where a

vascular complex forms rapidly during the first weeks of life

(Saint-GeniezandD’Amore, 2004).Consistentwitha role inactive

vessel growth, the endothelial cells of most vessels within the

Ccm2LGFP retina wereGFP+ at P6 (Figures 2S and 2T). However,

the tip cells that lead retinal vascular growth were mostly GFP�(Figure 2T, discussed further below). Despite a spatial associa-

tion between Ccm2LGFP expression and nascent vessels, GFP+

endothelial cells did not exhibit an increase in BrdU uptake

(Figure S1), indicating that expression of Ccm2L does not corre-

late directly with endothelial proliferation. These studies reveal

that Ccm2L expression is highly specific and restricted to endo-

thelial cells of the growing heart and vessels in a pattern con-

sistent with a role in active angiogenesis and cardiogenesis.

Heg and Ccm2L Function in a Pathway Requiredfor Cardiac GrowthTo test the requirement for CCM2L in vivo, we generated

Ccm2L�/� mice by intercrossing Ccm2LGFP/+ mice. Quantitative

FLAG-CCM2 added are shown below. Note that addition of FLAG-CCM2L or

HA-CCM2 to a similar degree. (G) The level of MYC-CCM1 coimmunoprecipita

was measured as in (F). (H) The level of MYC-CCM1 coimmunoprecipitated w

CCM2L-L306R.

Develop

real-time PCR (qPCR) and RT-PCR analysis of Ccm2LGFP/GFP

embryos and tissues revealed that the Ccm2LGFP allele fails to

express any Ccm2L mRNA 30 of exon 1 and (Figure S1) and is

therefore a null allele. Ccm2L�/� mice were born in normal

numbers from Ccm2L+/� intercrosses and grew to maturity

without overt phenotypes on mixed SV129J;C57Bl/6, 100%

SV129J, and 100% C57Bl/6 genetic backgrounds (Table S1).

Thus, unlike CCM2, CCM2L is not required for mouse cardiovas-

cular development.

Like Ccm2L, Heg is strongly expressed in endocardial and

endothelial cells (Kleaveland et al., 2009). HEG-deficient mice

exhibit predominantly postnatal cardiovascular phenotypes,

including thinning of the myocardial wall, which can result in

cardiac rupture and death (Kleaveland et al., 2009). In contrast,

CCM1-deficient and CCM2-deficient mouse embryos die

by E9.5 because of defective branchial arch artery lumen forma-

tion and an inability to circulate blood, a defect reproduced

by endothelial-specific loss of CCM2 (Boulday et al., 2009;

Whitehead et al., 2004, 2009). This phenotype is also observed

in Heg�/�;Ccm2+/� embryos (Kleaveland et al., 2009),

demonstrating that HEG and CCM2 function in a common

pathway and that loss of HEG sensitizes this pathway to partial

loss of CCM2. Because CCM2L associates with HEG

and CCM1, we next generated animals lacking both Heg

and Ccm2L to determine whether and how CCM2L might

interact with the known CCM signaling pathway. Analysis of

116 progeny of Heg+/�;Ccm2L+/� intercrosses revealed no

livebornHeg�/�;Ccm2L�/� animals (p < 0.01).Heg�/�;Ccm2L+/�

and Heg+/�;Ccm2L�/� animals survived to birth, but all

Heg�/�;Ccm2L�/� embryos died prior to E11.5 (Table S2). In

contrast to Ccm2�/� and Heg�/�;Ccm2+/� embryos, however,

Heg�/�;Ccm2L�/� embryos were viable at E9.5, developed

patent branchial arch arteries, and had normal blood circulation

at this time point (Figures 3A–3F and Table S2). By E10.5Heg�/�;Ccm2L�/� embryos exhibited severe myocardial thinning,

reduced ventricular trabeculation, and dilated atria (Figures

3G–3I). High-frequency ultrasound revealed a marked reduction

in the systolic fractional shortening of the left ventricle in E10.5

Heg�/�;Ccm2L�/� embryos (Figure 3L), confirming that

Heg�/�;Ccm2L�/�embryos die of heart failure. Analysis of

BrdU uptake revealed reduced proliferation of myocardial but

not endocardial cells in the hearts of E10.5 Heg�/�;Ccm2L�/�

embryos (Figure 3M). Thus, Heg and Ccm2L operate in a

common pathway in vivo that supports cardiac growth, a role

distinct from that of HEG-CCM2 signaling and consistent with

the expression pattern of Ccm2L.

HEG-CCM2L Signaling Regulates Endocardial GrowthFactor ExpressionBecause HEG and CCM2L physically associate and are ex-

pressed strongly in endocardial cells, the reduced myocardial

proliferation observed inHeg�/�;Ccm2L�/� embryos suggested

that HEG-CCM2L signaling might regulate the endocardial

production of growth factors required to support myocardial

FLAG-CCM2 reduce the level of CCM1 that is coimmunoprecipitated with

ted with HA-CCM2L in the presence of either FLAG-CCM2L or FLAG-CCM2

ith HA-CCM2 is reduced in the presence of FLAG-CCM2L but not FLAG-


0

10

20

30

Per

cent

age

of G

FP

+

End

ocar

dial

Cel

ls (

%)

E9.5E10

.5

E11.5E12

.5

E10.5 E11.5 E12.5 E15.5

G H I J K

P

L M N O O’

Q

da cv

nt

da

ntR S T

a

ao

B

Cao

a v

D

Ea v

ao

A F

0

300

600

900TIE2

CCM3CCM2LCCM2CCM1

HMVEC

HMVEC-C

LEC

HPK

Figure 2. Ccm2L Is Expressed in Endothelial Cells that Participate in Active Cardiovascular Growth

(A–E) In situ hybridization reveals endothelial-specific expression of Ccm2L. Shown are sagittal sections of an E11.5 mouse embryo. Ccm2L signal is shown in

pink (arrows) and DAPI staining of cell nuclei in blue. ao, aorta; cv, cardinal vein; a, atrium; v, ventricle.

(F) Quantitative RT-PCR measurement of mRNA transcripts encoding CCM1, CCM2, CCM2L, CCM3, and the endothelial-specific control gene TIE2

in cultured human microvascular endothelial cells from the skin (HMVEC) and heart (HMVEC-C), dermal lymphatic endothelial cells (LEC), and human

primary keratinocytes (HPK) is shown. The black and red arrowheads indicate that no expression of TIE2 or CCM2L was detected in HPKs. n = 4; error bars

indicate SEM.

(G–K) Ccm2LGFP expression in the endocardium. Shown are low (above) and high (below) magnification images of cardiac trabeculae from E10.5–E15.5 hearts

following immunostaining for GFP (green) and the endothelial cell marker PECAM (red). (K) Quantitation ofCcm2LGFP-expressing endocardial cells during mouse

cardiac development. n = 5; error bars indicate SEM.

(L–N) Expression of Ccm2LGFP in the dorsal aorta (da) and cardinal vein (cv) at E10.5.

Developmental Cell



Developmental Cell


growth at this time point (e.g., neuregulin and FGF family

members; Gassmann et al., 1995; Iwamoto et al., 2003; Lavine

et al., 2005; Lu et al., 2010). To test the requirement for HEG-

CCM2L signaling specifically in endothelial cells, we generated

mice carrying a conditional Heg allele using homologous recom-

bination in ES cells (Figure S2). Cre-mediated recombination of

the Hegfl allele results in deletion of exon 1 and creation of

a Heg allele that we have previously shown to be null (Figure S2;

Kleaveland et al., 2009). Analysis of 44 offspring of Tie2-

Cre;Ccm2L�/�Heg+/� 3 Ccm2L�/�Hegfl/fl matings revealed no

liveborn Tie2-Cre;Hegfl/�;Ccm2L�/� animals (p < 0.001). Tie2-

Cre;Hegfl/�;Ccm2L�/� animals exhibited embryonic lethality

at E11 associated with myocardial thinning, a phenotype

identical to that observed in Heg�/�;Ccm2L�/� animals (Figures

3J and 3K). These findings and the endothelial-specific expres-

sion pattern of Ccm2L indicate that HEG and CCM2L function

in a common endothelial pathway during cardiovascular

development.

To determine if HEG-CCM2L signaling regulates endocardial

growth factor expression, qPCR was performed to measure the

levels of Neuregulin, Fgf9, Fgf12, and Fgf16 in the hearts of

mice lacking CCM2L, HEG, or both HEG and CCM2L. Although

CCM2L-deficient embryos exhibited normal cardiac growth,

significantly reduced levels of Fgf12 and Fgf16 but not Fgf9

or Neuregulin mRNA were detected in E10.5 Ccm2L�/� hearts

at this time point (Figure 3N). At E9.5, a time point prior to any

detectable cardiac phenotype, when compared to Heg+/�;Ccm2L+/� littermates Heg�/�;Ccm2L+/�, Heg+/�;Ccm2L�/�,and Heg�/�;Ccm2L�/� hearts exhibited a graded loss of

expression of Fgf16, a gene required to stimulate myocardial

growth at this time point (Lavine et al., 2005; Lu et al., 2008)

(Figure 3O). Consistent with these loss-of-function findings

in vivo, adenoviral overexpression of CCM2L in endothelial cells

in vitro conferred an increase in FGF12 and FGF16 expression

(Figure 3P). These studies demonstrate that HEG-CCM2L

signaling regulates the expression of myocardial growth factors

by endothelial cells in the E9.5–E10.5 heart and explain

why combined deficiency of HEG and CCM2L is lethal at this

time point.

Ccm2L and Ccm2 Play Opposing Roles duringCardiovascular DevelopmentThe studies described above and published studies (Boulday

et al., 2009; Whitehead et al., 2009) establish that HEG,

CCM2, and CCM2L all function in endothelial cells during

cardiovascular development. The observations that (1)

CCM2L competes with CCM2 for CCM1 binding but does not

bind CCM3, (2) Ccm2L is expressed in a dynamic manner in

endothelial cells, and (3) Heg�/�;Ccm2L�/� embryos exhibit

a cardiovascular phenotype distinct from that of Ccm2�/�

and Heg�/�;Ccm2+/� embryos suggested either that CCM2L

and CCM2 operate in discrete endothelial signaling path-

ways downstream of HEG or that these two pathways may

(O–T) Ccm2LGFP is expressed in the endothelial cells of nascent vessels. Ccm2LG

vessels (O), such as those that invade the neural tube (nt) at E10.5–E11.5 (P an

neonatal retina (S and T). Scale bars indicate 20 mm unless otherwise indicated.

See also Figure S1 and Table S1.

Develop

compete with each other in a dynamic manner determined by

the relative levels of CCM2L and CCM2. To determine if

CCM2 and CCM2L function in discrete or intersecting path-

ways, we next performed genetic experiments to test the rela-

tionship between Ccm2 and Ccm2L during cardiovascular

development.

To test for functional redundancy between CCM2 and

CCM2L, we first generated Ccm2L�/�;Ccm2+/� compound

mutant animals. Ccm2L�/�;Ccm2+/� mice were born in normal

numbers and appeared healthy and fertile (data not shown),

suggesting that CCM2 and CCM2L are not redundant in

function. To further address whether CCM2 and CCM2L are

functionally redundant in vivo, we performed complementation

experiments in zebrafish embryos. The cardiac phenotype

conferred by loss of ccm2 in zebrafish embryos can be

efficiently rescued by the injection of wild-type ccm2 cRNA

(>90% rescue) but not by cRNA encoding a ccm2 PTB domain

mutant (Kleaveland et al., 2009). To determine if ccm2L

can compensate for loss of ccm2 in vivo, we coinjected ccm2

or ccm2L cRNAs with ccm2 morpholinos into zebrafish

embryos. cRNA encoding ccm2 but not ccm2L efficiently

rescued the cardiovascular phenotype of ccm2-morphant

embryos (Figures 4A–4E), despite successful expression of

ccm2L protein (data not shown). These studies, and the finding

that CCM2 but not CCM2L can bind CCM3, indicate that CCM2

and CCM2L do not play functionally redundant roles down-

stream of HEG.

Biochemical studies revealed that CCM2L and CCM2

compete for binding to CCM1 (Figures 1F and 1G), suggesting

that expression of CCM2L could reduce CCM2 signaling in

endothelial cells. To determine if CCM2L might modulate

signaling by CCM2, we next tested the effect of changes in the

balance of Ccm2 and Ccm2L gene dosage on cardiovascular

development by examining the effect of loss of a Ccm2 allele

on Heg-Ccm2L compound mutants and vice versa. Although

allHeg�/�;Ccm2L�/� embryos and allHeg�/�;Ccm2+/� embryos

died prior to E12 because of defects in heart and branchial

arch artery development, respectively, approximately half of

Heg�/�;Ccm2L�/�;Ccm2+/� animals generated by Heg+/�;Ccm2L�/� 3 Heg+/�;Ccm2L�/�;Ccm2+/� matings survived

(Figures 4F–4L, p < 0.05). Matings between Heg+/�;Ccm2L�/�;Ccm2+/� animals and surviving Heg�/�;Ccm2L�/�;Ccm2+/� animals confirmed that all Heg�/�;Ccm2L�/�;Ccm2+/+

offspring died in utero, whereas half of Heg�/�;Ccm2L�/�;Ccm2+/� offspring exhibited normal cardiovascular develop-

ment (Figures 4F–4K; Table S3, p < 0.01). Finally, intercrosses

of surviving Heg�/�;Ccm2L�/�;Ccm2+/� animals revealed that

all liveborn offspring were Heg�/�;Ccm2L�/�;Ccm2+/�, whereas

all Heg�/�;Ccm2L�/�;Ccm2+/+ and all Heg�/�;Ccm2L�/�;Ccm2�/� embryos died in utero (data not shown, p < 0.01).

Thus, the survival of Heg�/�;Ccm2L�/�;Ccm2+/� animals is an

effect of Ccm2 versus Ccm2L gene dosage and not selection

for a favorable background strain. Finally, to further test whether

FP was frequently detected in nonlumenized endothelial extensions of existing

d Q), and microvasculature (R), and in newly formed microvasculature of the


1010

Vel

ocity

(cm

/s)

0

-10

0 1000 Time (ms)

0

20

40

60

F

ract

iona

lS

hort

enin

g (%

)

*

0

102030

40

50

BrD

U+

Cel

lsin

Ven

tric

le (

%)

Total EC MC

* *

A DB E

C0

-10

10

Vel

ocity

(cm

/s)

0 1000 Time (ms)

F

L

G H I

G’ H’ I’

M

BAA BAA

Ccm2L Heg-/--/-

Ccm2L-/-

Ccm2L-/-

Ccm2L Heg-/--/-Ccm

2L-/-

Ccm2L

Heg

-/-

-/-

Ccm2L-/-

Ccm2L Heg-/--/- Ccm2L Hegfl/--/- Ccm2L Hegfl/--/-Tie2Cre+

0

0.25

0.50

0.75

1.00

1.25

CC

M2L

-/-:

CC

M2L

+/+

H

eart

Exp

ress

ion

Nrg1

Fgf9Fgf

12Fgf

16

N

0

0.50

1.00

1.50

Hea

rt E

xpre

ssio

n as

fold

of C

CM

2L+

/-H

eg+

/-

Nrg1 Fgf9 Fgf12 Fgf16

O

0

5

10

15

20

25

Ad-

Ccm

2L:A

d--L

acZ

Exp

ress

ion

inH

MV

EC

s

Fgf9Fgf

12Fgf

16

P

Ccm2L Heg+/-+/-

Ccm2L Heg+/--/-Ccm2L Heg-/-+/-

Ccm2L Heg-/--/-

1.14

0.79

0.440.30

1.11

19.95

4.68

J

J’

K

K’

*

P<0.01

*

*

**

Figure 3. HEG-CCM2L Signaling Is Required for Endocardial Growth Factor Expression during Cardiac Development

(A–F) Ccm2L�/�; Heg�/� embryos exhibit patent branchial arch arteries and normal blood circulation. E9.5 Ccm2L�/� and Ccm2L�/�; Heg�/� embryos

are shown. Blood can be visualized in the great vessels (A and D) and patent branchial arch arteries (BAAs) demonstrated by hematoxylin and eosin

(H&E) staining of transverse sections (B and E). Doppler ultrasound was used to detect normal systolic flow in the aorta of Ccm2L�/�; Heg�/� embryos

(C and F).

(G–I) Ccm2L�/�;Heg�/� embryos exhibit cardiac thinning, reduced trabeculation, and heart failure after E10.5. H&E stained transverse sections of E10.5

Ccm2L�/� and Ccm2L�/�; Heg�/� embryos are shown, with magnification of the boxed regions below. In addition to marked thinning of the heart wall (arrows,

H’ and I’) and reduced ventricular trabeculation, Ccm2L�/�; Heg�/� embryos develop dilated atria, a sign of cardiac failure.

(J and K) HEG-CCM2 signaling is required in the endocardium for myocardial growth. H&E stained transverse sections of E10.5 Ccm2L�/�;Hegfl/� and

Ccm2L�/�;Tie2-Cre;Hegfl/� embryos are shown, with magnification of the boxed regions below (J’ and K’).

(L) Ccm2L�/�;Heg�/� embryo hearts exhibit reduced systolic function. Fractional shortening of littermate embryo hearts was measured in utero at E10.5 with

high-frequency ultrasound. n = 4.

(M) Ccm2L�/�; Heg�/� embryo hearts have reduced BrdU uptake in myocardial, but not endocardial, cells at E10.5. n = 3. Scale bars indicate 50 mm.

Developmental Cell



F

Em

bryo

s w

ithD

ilate

d H

eart

(%

)

0

20

40

80

60

100

ccm2MO - + + +ccm2 - - + -

ccm2L - - - +

H

HegCcm2Ccm2L

G

I J K

-/-+/++/-

-/-+/-+/-

-/-+/--/-

BAA BAA

ccm2-MOControl

ccm2-MOccm2 cRNA

ccm2-MOccm2L cRNA

C

E

D

A B

Heg +/+ +/+ +/- +/- -/- -/-Ccm2 +/+ +/- +/+ +/- +/+ +/-Ccm2L -/- -/- -/- -/- -/- -/-

Obs

erve

d/E

xpec

ted

(%)

0

50

100

150L

Tie2Cre - + - +Heg fl/+ fl/+ fl/- fl/-Ccm2 fl/+ fl/+ fl/+ fl/+Ccm2L -/- -/- -/- -/-

0

50

100

150M

Obs

erve

d/E

xpec

ted

(%)

Figure 4. Ccm2 and Ccm2L Play Opposing

Rather than Redundant Roles during

Cardiovascular Development

(A–E) ccm2L does not rescue loss of ccm2 in

zebrafish embryos in vivo. Zebrafish embryos

were injected with control morpholino alone (A),

morpholino directed against ccm2 alone (B), or

with ccm2 cRNA (C) or with ccm2L cRNA (D), and

the presence of a dilated heart (red circle) pheno-

type scored 48 hpf. ccm2 cRNA conferred highly

efficient (>90%) rescue of the ccm2 morphant

phenotype, but ccm2L cRNA did not (E). n = 6;

error bars indicate SEM.

(F–K) Loss of Ccm2L rescues loss of Ccm2 during

cardiovascular development. E9.5 embryos lack-

ing different numbers of Ccm2 and Ccm2L alleles

were generated on a Heg�/� background to test

for genetic interaction. Loss ofCcm2L rescued the

defect in branchial arch artery (BAA) lumenization

and embryonic lethality observed with loss of

Ccm2 (I–K). White scale bars indicate 500 mm;

black scale bars indicate 100 mm.

(L) The ratio of observed/expected offspring

of Ccm2L�/�Heg+/� 3 Ccm2L�/�Heg+/�Ccm2+/�

matings at P14 is shown.

(M) The ratio of observed/expected offspring of

Tie2-Cre;Ccm2L�/�Heg+/� 3 Ccm2L�/�Hegfl/fl

Ccm2fl/fl matings at P14 is shown. Red lettering

and arrows indicate combinatorial lethality; green

lettering and arrows indicate combinatorial rescue

of lethality.


Developmental Cell


HEG, CCM2, and CCM2L interact specifically in endothelial

cells, we crossed Tie2-Cre;Heg+/�;Ccm2L�/� and Hegfl/fl;

Ccm2L�/�;Ccm2fl/fl animals (Figure S3). Although all Tie2-Cre;

Hegfl/�;Ccm2L�/� animals died before E11 because of cardiac

failure (Figures 3K and 4L), liveborn Tie2-Cre;Hegfl/�;Ccm2L�/�;Ccm2fl/+ animals were observed, demonstrating rescue with

endothelial-specific loss of CCM2 (Figure 4M). These genetic

studies identify an endothelial cell autonomous signaling

pathway and reveal that loss of CCM2L can compensate for

loss of CCM2 during development and vice versa. CCM2L and

CCM2 therefore play roles in opposing pathways downstream

of HEG in endothelial cells in vivo, and an appropriate balance

of CCM2 andCCM2L expression is critical at multiple time points

during cardiovascular development.

CCM2L Inhibits Endothelial Lumen Formationand Loosens Endothelial Cell Junctions In VitroEndothelial loss of CCM2 prevents lumen formation in collagen

gels in vitro and results in a nonlumenized branchial arch artery

(N) Ccm2L�/� embryo hearts exhibit reduced Fgf12 and Fgf16 mRNA expressio

qPCR and represented as the ratio of Ccm2L�/�:Ccm2L+/+ expression. n = 5 for

(O) Ccm2L�/�; Heg�/� embryo hearts exhibit severely reduced levels of Fgf16 mR

and represented as the ratio of the indicated genotypes to Ccm2L+/�; Heg+/� ex

(P) CCM2L drives expression of FGF12 and FGF16 in cultured endothelial cells

encoding Ccm2L or LacZ and the indicated mRNAs measured using qPCR after 4

(L–P) indicate SEM.


Develop

in vivo (Kleaveland et al., 2009; Whitehead et al., 2009). The

observation that loss of CCM2L can rescue branchial arch artery

lumenization in Heg�/�;Ccm2+/� embryos suggested that

CCM2L may also oppose the known role of CCM2 in lumen

formation. Ccm2LGFP expression is only detected in the endo-

thelial cells of growing cardiovascular organs in vivo, and

Ccm2L expression in cultured endothelial cells was also

extremely lowwhenmeasured by RT-PCR (Figure S4). We there-

fore used adenoviral vectors to express CCM2L or the control

proteins b-galactosidase (LacZ) or GFP in cultured endothelial

cells. Expression of CCM2L, but not LacZ or GFP, significantly

inhibited lumen formation by cultured HUVECs (Figures 5A and

5B; Movie S1). Conversely, siRNA directed against CCM2L

(Figure S4) conferred an increase in endothelial lumen formation

(Figures 5C and 5D).

In vivo and in vitro studies have established that HEG-CCM2

signaling also positively regulates endothelial cell junctions,

in part by reducing RHO activity and expression level (Borikova

et al., 2010; Glading et al., 2007; Kleaveland et al., 2009;

n at E9.5. mRNA levels of the indicated growth factors were measured using

each genotype.

NA. mRNA levels of the indicated growth factors were measured using qPCR

pression. n = 4 for each genotype.

. Human microvascular endothelial cells were exposed to adenoviral vectors

8 hr. n = 3 for FGF9; n = 10 for both FGF12 and FGF16. *p < 0.01. Error bars in


Fol

d of

Con

trol

0

3

2

1

RHOA

CDC42RAC1

*Ad-LacZAd-CCM2L

RHOA

ACTIN

RHOA-GTP

Ad-LacZ + -Ad-CCM2L - +

A B

Bas

al T

EE

R (

Ohm

s)

0

1500

1000

500

2500

2000

Ad-La

cZ

Ad-CCM

2L

*

C D

Lum

en a

rea

(% c

ontr

ol)

siRNA:

Contro

l

CCM2L

G

Control siRNA CCM2L siRNA

Fol

d of

Con

trol

0

1.5

1.0

0.5

RHOA

CDC42RAC1

si-Controlsi-CCM2L

CCM2L

Ad-

GF

PA

d-C

CM

2L 0

60

40

20

80

120

100

Ad-GFP

Ad- C

CM2LLL

P<0.01

E F

H

0

150100

50

200

250

Lum

en a

rea

(%

con

trol

)

P<0.05

ACTIN

RHOA

siRNA:Con

trol I

CCM2L

RHOA-GTP

CCM2L

Figure 5. CCM2L Opposes CCM2 in the

Regulation of Lumen Formation and Endo-

thelial Junctions

(A and B) Expression of CCM2L inhibits endothelial

lumen formation in vitro. Adenoviral vectors were

used to express CCM2L or the control proteins

GFP or LacZ in human umbilical vein endothelial

cells (HUVECs) and lumen formation measured in

a 3D collagen gel. Shown are images of endothelial

cells in representative collagen gels at the indi-

cated time points (A) and the percentage of lumen

area relative to control vector-treated cells (B).

n = 20 collagen gels for the adeno-GFP controlled

experiment and n = 10 for the adeno-CCM2L

experiment. Scale bar is 50 mM. Arrowheads indi-

cate developing multicellular lumens, and arrows

indicate the final border of the lumen structure.

(C and D) Loss of CCM2L accelerates endothelial

lumen formation in vitro. HUVEC lumen formation

was measured following treatment with siRNA

directed against CCM2L or control siRNA. Shown

are images of endothelial cells in representative

collagen gels at 24 hr (C) and the percentage of

lumen area relative to control vector-treated cells

(D). n = 40 collagen gels analyzed for each group.

Arrows indicate lumenized structures.

(E) CCM2L expression reduces microvascular

endothelial cell trans-endothelial electrical resis-

tance (TEER). Microvascular endothelial cells were

treated with adenovirus to express the control

protein b-galactosidase (LacZ) or FLAG-CCM2L

and TEERmeasured on confluent cell monolayers.

n = 4. Asterisk indicates p < 0.01.

(F) CCM2L expression increases the levels of

RHO-A activation and total protein. Shown are

representative immunoblots to detect RHOA-GTP,

RHOA, FLAG-CCM2L, and ACTIN in the micro-

vascular endothelial cell lysate after exposure to

adeno-b-galactosidase and adeno-CCM2L. n = 5

for total RHOA, n = 2 for RHOA-GTP.

(G) CCM2L expression raises the levels of RHOA

but not RAC1 or CDC42 mRNA as detected by

qPCR. n = 5. Asterisk indicates p < 0.01.

(H) Loss of CCM2L does not significantly alter the

level of RHOA activation or protein expression in

microvascular endothelial cells. The blot shown is

representative of two independent experiments.

(I) Loss of CCM2L does not significantly alter expression of RHOA, RAC1, or CDC42 in microvascular endothelial cells. mRNA levels were detected

using RT-qPCR. n = 4; error bars in (B), (D), (E), (G), and (I) indicate SEM.

See also Figure S4.

Developmental Cell


Whitehead et al., 2009; Zheng et al., 2010). Expression of

CCM2L significantly reduced transendothelial resistance

(TEER, a measure of endothelial junction tightness) (Figure 5E),

an effect similar to that previously observed with loss of CCM2

or CCM3 (Whitehead et al., 2009; Zheng et al., 2010). Forced

expression of CCM2L also raised the levels of both activated

RHOA (RHOA-GTP), total RHOA protein and RHOA, but not

RAC1 or CDC42, transcripts in microvascular endothelial cells

(Figures 5F and 5G). Reduction of the low basal level of

CCM2L levels in these cells did not significantly alter the level

of activated or total RHOA protein or RHOA mRNA (Figures 5H

and 5I). These findings demonstrate that CCM2L and CCM2

have opposing effects in endothelial cells in vitro as well as

in vivo.


Loss of CCM2L Retards Tumor Growth and WoundHealingThe studies described above suggested that CCM2L may play

an important role in promoting vascular growth despite the fact

that it is not required during cardiovascular development. To

further test a potential angiogenic role for CCM2L, we next

examined tumor growth and wound healing, processes known

to require robust vessel growth (Conway et al., 2001; Lyden

et al., 2001). To test tumor growth, 4 3 105 Lewis Lung Carci-

noma (LLC) cells were injected into the flank of Ccm2L�/� mice

and littermate controls. LLC tumor growth was markedly

impaired in Ccm2L�/� mice relative to wild-type littermates,

and tumors grown in Ccm2L�/� animals appeared pale in

comparison with those in control animals (Figures 6A–6C).

r Inc.

0

100

200

300

400

500

5 7 9 11 13 15

600

Tum

or V

olum

e (m

m3)

Days After Injection

100

75

50

25

1 3 5 70

* **

Wou

nd A

rea

(%)

Days After WoundingCcm2LCcm2L

A B

C

D E

Ccm2LCcm2L

+/+

-/-

+/+

Sur

viva

l (%

) 100

80

60

20

0

40

0 20 40 60 80Days After Injection

Ccm2L (n = 17)

Ccm2L (n = 26)

+/+

-/-

-/-

Ccm2L

Ccm2L+/+

-/-

Ccm2LCcm2L0

200

400

600

800

1000

Tum

or W

eigh

t (m

g)

+/+ -/-

FG

* ** ** **

**

0

200

400

7 9 11 13 15

800

Tum

or V

olum

e (m

m3)


600

***

WT BM to Ccm2L-/-WT BM to Ccm2L+/+

0

200

400

7 9 11 13 15 17

800

Tum

or V

olum

e (m

m3)


600

Ccm2L BM to WTCcm2L BM to WT

+/+

-/-

Figure 6. CCM2L Is Required for Rapid

Tumor Growth and Wound Healing In Vivo

(A–C) Growth of Lewis Lung Carcinoma (LLC) cells

is severely retarded in CCM2L-deficient mice.

Four hundred thousand LLC cells were injected

into the flank of mature mice and tumor volume

measured at the indicated times after injection.

Tumor weight was measured after harvest at

15 days. n = 10. Representative tumors are shown.

Note the small size and lack of vascularity in

tumors harvested from Ccm2L�/� animals.

(D) Ccm2L�/� mice are protected from death after

tail vein injection of LLC cells. Survival curves

following tail vein injection of 5 3 105 LLC cells

are shown.

(E and F) Growth of Lewis Lung Carcinoma (LLC)

cells is inhibited in irradiated Ccm2L�/� mice re-

constituted with wild-type bone morrow (E, n = 10)

but not irradiated wild-type mice reconstituted

with Ccm2L�/� bone morrow (F, n = 12).

(G) Wound healing is retarded in Ccm2L�/� mice.

Eight millimeter full thickness wounds were made

in wild-type and Ccm2L�/� animals and the

percentage of the original wounded area calcu-

lated at the indicated time points after wounding.

n = 8; error bars in (A), (B), (E), (F), and (G)

indicate SEM.

Developmental Cell


Ccm2L�/� mice were also protected from lethality because of

pulmonary metastases following tail vein injection of 5 3 105

LLC cells (Figure 6D). These findings suggested that endothelial

CCM2L is required for optimal tumor angiogenesis, but

Ccm2LGFP expression could not be detected in the tumors of

Ccm2L+/� animals. To further address the source of CCM2L,

LLC cells were injected into lethally irradiatedCcm2L�/� animals

that were reconstituted with wild-type bone marrow (Figure 6E)

and wild-type mice reconstituted with Ccm2L�/� bone marrow

(Figure 6F). These studies demonstrated a requirement for

CCM2L in nonhematopoietic cells, consistent with low level

expression of CCM2L in the endothelium of tumor vessels. To

test the role of CCM2L in a more physiologic angiogenic

process, we compared wound healing following full thickness

skin punch biopsy in Ccm2L�/� mice and wild-type littermates.

Wound healing was significantly delayed in Ccm2L�/� mice

relative to controls (Figure 6E). These findings suggest that

Developmental Cell 23, 342–355

although CCM2L facilitates rapid angio-

genic responses in postnatal animals.

DISCUSSION

CCM2L Promotes VascularRemodeling by Blocking CanonicalCCM SignalsDiseases such as ischemia and cancer

are characterized by inadequate or

excessive vascular growth. In the former,

vessels remain excessively stable and fail

to sprout new vessels to feed blood-

starved tissues. In the latter, there is

excess vascular permeability and new

vessel formation that fuels tumor growth and inflammation.

Understanding the molecular signals that balance vascular

stability and growth is essential to devise effective angiogenic

therapies. The CCM pathway has recently been identified as

a positive regulator of endothelial cell junctions and vessel

stability in both developing and mature animals (Kleaveland

et al., 2009; Stockton et al., 2010; Whitehead et al., 2009).

Ccm2 is expressed at low levels in endothelial cells, and

increased permeability in Ccm2+/� mice (Stockton et al., 2010)

is consistent with a role for CCM2 in generating tonic stabilizing

signals in the cardiovascular system. In the present study we

identify CCM2L as an endothelial-specific modulator of CCM

signaling that is linked to cardiovascular growth. Our studies

suggest that CCM2L both opposes CCM2-mediated endothelial

and vascular stability and activates endothelial growth factor

expression in the heart. These findings are consistent with

a model in which dynamic expression of CCM2L functions as

, August 14, 2012 ª2012 Elsevier Inc. 351

HEGNPxYCCM1

CCM3

CCM2LPTB

PTB

CCM2

ECM

HEG NPxYCCM1

CCM3

PTBCCM2

ECM

CCM2L+

Endothelium

Lack of CCM2L Tonic CCM signaling& cell junction stability

Endothelium

CCM2L expression CCM3 uncoupling &cell junction release Vessel growth

Cell junction

Cell junction

RHO

RHO

CCM2LPTB

HEG NPxYCCM1

Cardiac jelly

Endothelium

CCM2L expression endocardial growth factor expression Heart growth

FGF-16FGF-12

A

B

C

Cardiomyocyte

Vessel stability

Figure 7. Model of CCM2L Function during

Cardiovascular Growth

(A) In the absence of CCM2L, constitutive

expression of CCM2 confers tonic positive regu-

lation of endothelial junctions and vessel stability.

(B) In response to as yet unidentified angiogenic

signals, CCM2L is expressed in a restricted

number of endothelial cells in actively growing

cardiovascular organs. CCM2L competes with

CCM2 for CCM1 and uncouples CCM1 from

CCM3 to break the tonic CCM signal. In the

absence of CCM2-CCM3 signaling, CCM2L-

expressing endothelial cells are able to uncouple

from neighboring endothelial cells in preparation

for cellular proliferation and migration.

(C) HEG-CCM2L signaling in the developing heart

drives endocardial expression of growth factors

and myocardial proliferation.

Developmental Cell


a molecular means of converting a tightly connected, quiescent

endothelial cell to a one that is disconnected and actively

participating in vessel or cardiac growth (Figure 7). CCM2L upre-

gulation therefore provides a simple and elegant molecular

mechanism by which stabilizing signals are turned off and

growth signals turned on in a coordinated manner via changes

in a single endothelial cell pathway.

Tight Regulation of CCM2L Expression Reflects Its Rolein Balancing Vessel Stability versus Growth through theCCM PathwayComparison of CCM2L expression with that of CCM2 reveals

important differences and similarities. Studies of Ccm2LacZ

animals reveal broad Ccm2 expression in vivo with no detect-

able vascular pattern or dynamic changes in gene expression

level (Kleaveland et al., 2009). In contrast to Ccm2, studies of

Ccm2LGFP animals reveal tight precise spatial and temporal

control of Ccm2L expression in vivo. Spatially, Ccm2LGFP is

entirely restricted to endothelial cells, consistent with a highly

specific role in regulating cardiovascular CCM signaling.

Temporally, Ccm2LGFP expression is very dynamic and highest

during peak cardiac and vessel growth (e.g., in the E11 heart

and neonatal retina). Like both Ccm1 and Ccm2, however,

Ccm2L expression is undetectable in a majority of endothelial

cells in vivo, even when deficient animals exhibit significant

cardiovascular phenotypes, such as defects in tumor angio-


genesis and wound healing. Our bio-

chemical, cellular, and genetic studies

suggest that a primary role of CCM2L

is to oppose CCM2 by competing for

binding to CCM1 and thereby dynami-

cally regulate CCM signaling. Dynamic

modulation of CCM signaling using this

mechanism is only possible if the relative

levels of these three proteins is in a

similar range so that endothelial cells

can transition rapidly between stable,

nonangiogenic states and unstable,

angiogenic states by altering CCM2L

expression. Consistent with such a

mechanism, we detected similar levels of mRNA transcripts

encoding CCM2L and other CCM proteins in cultured endothe-

lial cells (Figure 2F). Although CCM2L may also affect CCM

signaling through an as yet unidentified indirect mechanism,

these studies support a model in which endothelial CCM

signaling is mediated by low levels of CCM1, CCM2, and

CCM2L that are tightly balanced by expression of CCM2L to

regulate endothelial cell stability versus growth in the func-

tioning vascular network.

The Role of CCM2L in Angiogenesis May Be to Enablebut Not Stimulate Endothelial ProliferationThe in vivo phenotypes of CCM2L-deficient mice support

a role for CCM2L in angiogenesis. Most of the angiogenic

regulators identified to date that act on or within endothelial

cells have been shown to regulate endothelial proliferation.

Despite a correlation with active angiogenic states, however,

GFP+ endothelial cells in Ccm2LGFP animals are not more

BrdU+ than GFP� endothelial cells, we do not detect high

Ccm2LGFP expression in retinal tip cells, and Ccm2L levels in

cultured endothelial cells are not increased by VEGF (Fig-

ure S4B). Instead, our biochemical, cellular, and genetic

studies support a specific role for CCM2L in opposing the

stabilizing signals mediated by CCM2 and CCM3. These find-

ings are consistent with a model in which CCM2L enables

endothelial cells to respond effectively to growth factors such

Developmental Cell


as VEGF by releasing them from neighboring cells, but is not

a proliferative signal itself. For example, in the retina tip cells

arise by active competition among stalk cells for VEGF

signaling (Jakobsson et al., 2010). By the time an endothelial

cell assumes a tip cell position, it has already responded to

VEGF signals and has been released from the constraint of

other endothelial cells and perhaps downregulated Ccm2L

expression.

CCM2L Couples CCM Signaling to Cardiac GrowthOur studies identify two lethal loss of function phenotypes,

defective branchial arch artery formation in E9 Heg�/�;Ccm2+/� embryos and inadequate cardiac growth in E10.5

Heg�/�;Ccm2L�/� embryos, that can be reversed by further

loss of either CCM2L or CCM2 to restore the balance of these

two proteins. Defective branchial arch artery formation in

Ccm2�/� and Heg�/�;Ccm2+/� embryos occurs when endothe-

lial cells align correctly along the branchial arch but fail to asso-

ciate into a lumenized vessel that connects the heart to the

dorsal aorta (Boulday et al., 2009; Kleaveland et al., 2009; White-

head et al., 2009). This in vivo phenotype strongly resembles the

failure of CCM2-deficient endothelial cells to lumenize in

collagen gels in vitro (Whitehead et al., 2009) and is consistent

with the known role for CCM signaling in regulating endothelial

cell-cell association and lumen formation. In contrast, HEG-

CCM2L signaling is required for cardiac trabeculation and

myocardial growth at E10–E11 (Figure 3), and expression

studies of embryonic hearts lacking one or both of these

proteins as well as endothelial cells that overexpress CCM2L

support reduced endothelial expression of essential myocardial

growth factors, such as FGF16 (Lu et al., 2008), as the mecha-

nism for this phenotype. One possible explanation for the drop

in endocardial growth factor expression in these animals is

a failure of endocardial cells to disengage and expand because

of inadequate suppression of CCM2 signaling, but endocardial

proliferation is not reduced in these hearts (Figure 3M). These

findings therefore suggest that HEG-CCM2L signaling supports

cardiac growth through regulation of endothelial growth factor

expression. The observation that cardiac lethality in Heg�/�;Ccm2+/� embryos can be reversed by loss of one Ccm2 allele

is consistent with bidirectional competition and a mechanism

by which myocardial growth is coupled to endocardial growth

in the developing embryo. Further studies to test the role of

HEG-CCM2L signaling and identify the downstream effectors

by which it controls the expression of endocardial growth

factors are an important next step in these studies.

EXPERIMENTAL PROCEDURES

Mice

The Ccm2L gene was disrupted in SV129 ES cells by replacing the coding

portion of exon 1, intron 1, and exon 2 with a cassette expressing nuclear-

GFP using recombineering-based gene-targeting techniques. HEG-deficient

and CCM2-deficient mice have been described previously (Kleaveland et al.,

2009). The conditional Heg allele was generated using gene-targeting as

described in Figure S2. The conditional Ccm2 allele was generated using

gene-targeting as described in Figure S3. Tie2-Cre transgenic mice were

obtained from Jackson Research Laboratories (Bar Harbor, ME, USA). The

University of Pennsylvania Institutional Animal Care and Use Committee

approved all animal protocols.

Develop

Histology

Embryos at desired developmental stages were dissected and analyzed as

previously described (Kleaveland et al., 2009). The primers used to generate

a Ccm2L in situ hybridization probe are listed in the Supplemental Experi-

mental Procedures. The following antibodies were used for immunostaings:

goat anti-GFP (1:100, Abcam, Cambridge, UK), rat anti-Flk1 (1:50, BD Phar-

Mingen, Franklin Lakes, NJ, USA), rat anti-Pecam (1:500, BD PharMingen),

and mouse anti-BrDU (1:10, Hybridoma Bank, Iowa City, IA, USA).

In Vivo BrDU Incorporation Assay

Pregnant mice were injected intraperitoneally with 100 mg/g body weight

BrDU. The embryos were dissected 1.5 hr after the injection, fixed, and

embedded in paraffin. Paraffin sections were immunostained with anti-BrDU

and anti-PECAM antibodies and nuclei visualized with DAPI.

Tumor Xenograft, Bone Marrow Transplant, and Wound Healing

Studies

Three-month-old Ccm2L�/� and littermate control mice were injected with

43 105 Lewis Lung Carcinoma (LLC) cells subcutaneously on the flank. Start-

ing one week after injection, tumor size wasmeasured every two days. Tumors

were excised andweighed 15 days after injection. For radiation chimera exper-

iments bonemarrow was isolated fromCcm2L�/� or control littermate animals

and transplanted by retro-orbital injection into recipient Ccm2L�/� or control

littermate animals conditioned with a split lethal dose of 10 Gy irradation.

LLC cells were injected subcutaneously 8 weeks after transplantation. To

test the survival following lung metastasis, 3-month-old Ccm2L�/� and litter-

mate control mice were injected with 53 105 LLC cells via tail vein and survival

monitored daily. Full thickness wounds were made on the dorsum of 3-month-

old Ccm2L�/� and littermate control mice using an 8-mm-skin biopsy punch.

Wound dimensions were measured every other day and the wound area

calculated.

Fetal Ultrasound

Pregnant mice were lightly anesthetized with 1%–2% isoflurane and

trans-uterine embryonic ultrasound performed as previously described (Lee

et al., 2006). Left ventricular systolic function was estimated by the percent

change in fractional area (FAC%), which was derived using the formula:

(Vd-versus) / Vd 3 100. Doppler ultrasound was used to detect blood flow in

the dorsal aorta.

Zebrafish Studies

Tuebingen long-fin wild-type zebrafish were maintained with approval of the

Institutional Animal Care andUse Committee of the University of Pennsylvania.

Antisense morpholino oligonucleotides (Gene Tools) that interfere with the

splicing of ccm2 (Mably et al., 2003, 2006) were injected into the yolk of

one-cell stage embryos at a dose of 5 ng. To rescue the phenotype conferred

by ccm2 morpholinos, 100 pg of cRNA encoding ccm2 or ccm2L was coin-

jected with the ccm2 morpholino oligonucleotides.

Measurement of Trans-Endothelial Electrical Resistance

Human dermal microvascular endothelial cell (HMEC) barrier function was

assayed by measuring the resistance of a cell-covered electrode using an

ECIS instrument (Applied BioPhysic, Troy, NY, USA) as previously described

(Zheng et al., 2010). Cells were infected with adenovirus expressing b-galac-

tosidase or CCM2L at 5,000 gc/cell. TEER was measured 72 hr after infection.

Endothelial Lumen Formation

Endothelial cell lumen formation assays were performed in three-dimensional

(3D) collagen matrices, real-time movies acquired, and cultures fixed and

quantitated for lumen formation as described as described (Koh et al., 2008).

Immunoprecipitation, Immunoblotting, and RHO Activation Assays

Biochemical studies of epitope-tagged proteins heterologously expressed in

HEK293T cells were performed as previously described (Kleaveland et al.,

2009; Zheng et al., 2010). RHO activation was measured using a RHO Activa-

tion Assay Biochem Kit (Cytoskeleton, Denver, CO, USA). Anti-ACTIN staining

(1:5,000; Cell Signaling, Danvers, MA, USA) was used as a loading control.


Developmental Cell


Statistics

The p values were calculated using an unpaired two-tailed Student’s t test,

ANOVA, or chi-square analysis as indicated.

SUPPLEMENTAL INFORMATION

Supplemental Information includes four figures, three tables, Supplemental

Experimental Procedures, and one movie and can be found with this article

online at http://dx.doi.org/10.1016/j.devcel.2012.06.004.

ACKNOWLEDGMENTS

We thank the members of the Kahn laboratory for their thoughtful comments

during the course of this work and Katherine Speichinger and Matthew Davis

for their technical assistance. We would also like to thank M. Ginsberg and

J.J. Liu for generously providing us with HEG-IC and aIIb-IC beads. These

studies were supported by grants from the National Institute of Health (grants

R01HL094326 and R01HL102138 to M.L.K., R01HL059373 to G.E.D.,

and T32HL07971 to X.Z.) and the American Heart Association (grant

11SDG7430025 to X.Z.).

Received: November 22, 2011

Revised: April 25, 2012

Accepted: June 5, 2012

Published online: August 13, 2012

REFERENCES

Borikova, A.L., Dibble, C.F., Sciaky, N., Welch, C.M., Abell, A.N., Bencharit, S.,

and Johnson, G.L. (2010). Rho kinase inhibition rescues the endothelial cell

cerebral cavernous malformation phenotype. J. Biol. Chem. 285, 11760–

11764.

Boulday, G., Blecon, A., Petit, N., Chareyre, F., Garcia, L.A., Niwa-Kawakita,

M., Giovannini, M., and Tournier-Lasserve, E. (2009). Tissue-specific condi-

tional CCM2 knockout mice establish the essential role of endothelial CCM2

in angiogenesis: implications for human cerebral cavernous malformations.

Dis. Model. Mech. 2, 168–177.

Clatterbuck, R.E., Eberhart, C.G., Crain, B.J., and Rigamonti, D. (2001).

Ultrastructural and immunocytochemical evidence that an incompetent

blood-brain barrier is related to the pathophysiology of cavernous malforma-

tions. J. Neurol. Neurosurg. Psychiatry 71, 188–192.

Conway, E.M., Collen, D., and Carmeliet, P. (2001). Molecular mechanisms of

blood vessel growth. Cardiovasc. Res. 49, 507–521.

Dejana, E., Orsenigo, F., Molendini, C., Baluk, P., and McDonald, D.M. (2009).

Organization and signaling of endothelial cell-to-cell junctions in various

regions of the blood and lymphatic vascular trees. Cell Tissue Res. 335,

17–25.

Denier, C., Goutagny, S., Labauge, P., Krivosic, V., Arnoult, M., Cousin, A.,

Benabid, A.L., Comoy, J., Frerebeau, P., Gilbert, B., et al; Societe Francaise

de Neurochirurgie. (2004). Mutations within theMGC4607 gene cause cerebral

cavernous malformations. Am. J. Hum. Genet. 74, 326–337.

Gassmann, M., Casagranda, F., Orioli, D., Simon, H., Lai, C., Klein, R., and

Lemke, G. (1995). Aberrant neural and cardiac development in mice lacking

the ErbB4 neuregulin receptor. Nature 378, 390–394.

Glading, A., Han, J., Stockton, R.A., and Ginsberg, M.H. (2007). KRIT-1/CCM1

is a Rap1 effector that regulates endothelial cell cell junctions. J. Cell Biol. 179,

247–254.

Iwamoto, R., Yamazaki, S., Asakura, M., Takashima, S., Hasuwa, H., Miyado,

K., Adachi, S., Kitakaze, M., Hashimoto, K., Raab, G., et al. (2003). Heparin-

binding EGF-like growth factor and ErbB signaling is essential for heart

function. Proc. Natl. Acad. Sci. USA 100, 3221–3226.

Jakobsson, L., Franco, C.A., Bentley, K., Collins, R.T., Ponsioen, B., Aspalter,

I.M., Rosewell, I., Busse, M., Thurston, G., Medvinsky, A., et al. (2010).

Endothelial cells dynamically compete for the tip cell position during angio-

genic sprouting. Nat. Cell Biol. 12, 943–953.


Kleaveland, B., Zheng, X., Liu, J.J., Blum, Y., Tung, J.J., Zou, Z., Sweeney,

S.M., Chen, M., Guo, L., Lu, M.M., et al. (2009). Regulation of cardiovascular

development and integrity by the heart of glass-cerebral cavernous malforma-

tion protein pathway. Nat. Med. 15, 169–176.

Koh, W., Stratman, A.N., Sacharidou, A., and Davis, G.E. (2008). In vitro three

dimensional collagen matrix models of endothelial lumen formation during

vasculogenesis and angiogenesis. Methods Enzymol. 443, 83–101.

Lavine, K.J., Yu, K., White, A.C., Zhang, X., Smith, C., Partanen, J., and

Ornitz, D.M. (2005). Endocardial and epicardial derived FGF signals

regulate myocardial proliferation and differentiation in vivo. Dev. Cell 8,

85–95.

Lee, J.S., Yu, Q., Shin, J.T., Sebzda, E., Bertozzi, C., Chen, M., Mericko, P.,

Stadtfeld, M., Zhou, D., Cheng, L., et al. (2006). Klf2 is an essential regulator

of vascular hemodynamic forces in vivo. Dev. Cell 11, 845–857.

Lu, S.Y., Sheikh, F., Sheppard, P.C., Fresnoza, A., Duckworth, M.L., Detillieux,

K.A., and Cattini, P.A. (2008). FGF-16 is required for embryonic heart develop-

ment. Biochem. Biophys. Res. Commun. 373, 270–274.

Lu, S.Y., Jin, Y., Li, X., Sheppard, P., Bock, M.E., Sheikh, F., Duckworth, M.L.,

and Cattini, P.A. (2010). Embryonic survival and severity of cardiac and cranio-

facial defects are affected by genetic background in fibroblast growth factor-

16 null mice. DNA Cell Biol. 29, 407–415.

Lyden, D., Hattori, K., Dias, S., Costa, C., Blaikie, P., Butros, L., Chadburn, A.,

Heissig, B., Marks, W., Witte, L., et al. (2001). Impaired recruitment of bone-

marrow-derived endothelial and hematopoietic precursor cells blocks tumor

angiogenesis and growth. Nat. Med. 7, 1194–1201.

Mably, J.D., Mohideen, M.A., Burns, C.G., Chen, J.N., and Fishman, M.C.

(2003). Heart of glass regulates the concentric growth of the heart in zebrafish.

Curr. Biol. 13, 2138–2147.

Mably, J.D., Chuang, L.P., Serluca, F.C., Mohideen, M.A., Chen, J.N., and

Fishman, M.C. (2006). santa and valentine pattern concentric growth of

cardiac myocardium in the zebrafish. Development 133, 3139–3146.

Murohara, T., Horowitz, J.R., Silver, M., Tsurumi, Y., Chen, D., Sullivan, A., and

Isner, J.M. (1998). Vascular endothelial growth factor/vascular permeability

factor enhances vascular permeability via nitric oxide and prostacyclin.

Circulation 97, 99–107.

Potente, M., Gerhardt, H., and Carmeliet, P. (2011). Basic and therapeutic

aspects of angiogenesis. Cell 146, 873–887.

Risau, W. (1997). Mechanisms of angiogenesis. Nature 386, 671–674.

Saint-Geniez, M., and D’Amore, P.A. (2004). Development and pathology of

the hyaloid, choroidal and retinal vasculature. Int. J. Dev. Biol. 48, 1045–1058.

Senger, D.R., Galli, S.J., Dvorak, A.M., Perruzzi, C.A., Harvey, V.S., and

Dvorak, H.F. (1983). Tumor cells secrete a vascular permeability factor that

promotes accumulation of ascites fluid. Science 219, 983–985.

Stockton, R.A., Shenkar, R., Awad, I.A., and Ginsberg, M.H. (2010). Cerebral

cavernous malformations proteins inhibit Rho kinase to stabilize vascular

integrity. J. Exp. Med. 207, 881–896.

Stolt, P.C., Jeon, H., Song, H.K., Herz, J., Eck,M.J., and Blacklow, S.C. (2003).

Origins of peptide selectivity and phosphoinositide binding revealed by struc-

tures of disabled-1 PTB domain complexes. Structure 11, 569–579.

Uhlik, M.T., Temple, B., Bencharit, S., Kimple, A.J., Siderovski, D.P., and

Johnson, G.L. (2005). Structural and evolutionary division of phosphotyrosine

binding (PTB) domains. J. Mol. Biol. 345, 1–20.

Voss, K., Stahl, S., Schleider, E., Ullrich, S., Nickel, J., Mueller, T.D., and

Felbor, U. (2007). CCM3 interacts with CCM2 indicating common pathogen-

esis for cerebral cavernous malformations. Neurogenetics 8, 249–256.

Voss, K., Stahl, S., Hogan, B.M., Reinders, J., Schleider, E., Schulte-Merker,

S., and Felbor, U. (2009). Functional analyses of human and zebrafish

18-amino acid in-frame deletion pave the way for domain mapping of the cere-

bral cavernous malformation 3 protein. Hum. Mutat. 30, 1003–1011.

Whitehead, K.J., Plummer, N.W., Adams, J.A., Marchuk, D.A., and Li, D.Y.

(2004). Ccm1 is required for arterial morphogenesis: implications

for the etiology of human cavernous malformations. Development 131,

1437–1448.

r Inc.


Developmental Cell


Whitehead, K.J., Chan, A.C., Navankasattusas, S., Koh, W., London, N.R.,

Ling, J., Mayo, A.H., Drakos, S.G., Jones, C.A., Zhu,W., et al. (2009). The cere-

bral cavernousmalformation signaling pathway promotes vascular integrity via

Rho GTPases. Nat. Med. 15, 177–184.

Zawistowski, J.S., Stalheim, L., Uhlik, M.T., Abell, A.N., Ancrile, B.B., Johnson,

G.L., and Marchuk, D.A. (2005). CCM1 and CCM2 protein interactions in cell

Develop

signaling: implications for cerebral cavernous malformations pathogenesis.

Hum. Mol. Genet. 14, 2521–2531.

Zheng, X., Xu, C., Di Lorenzo, A., Kleaveland, B., Zou, Z., Seiler, C., Chen, M.,

Cheng, L., Xiao, J., He, J., et al. (2010). CCM3 signaling through sterile 20-like

kinases plays an essential role during zebrafish cardiovascular development

and cerebral cavernous malformations. J. Clin. Invest. 120, 2795–2804.


Dynamic Regulation of the Cerebral Cavernous Malformation Pathway Controls Vascular Stability and...

Documents

Transcript of Dynamic Regulation of the Cerebral Cavernous Malformation Pathway Controls Vascular Stability and...